In Prisma Cloud, when anomalies are detected and the security team chooses to Relearn on a specific image, the existing behavioral model for that image is not deleted. Instead, the system retains the model and enters a new learning period, during which it observes the behavior of the container based on the image. If new behaviors are observed during this period, they are added to the existing model, thereby refining and updating the model to reflect the current operational profile of the container. This approach allows for dynamic adaptation to changes in container behavior while preserving the valuable insights and patterns already established in the model. The Relearn function is part of Prisma Cloud's adaptive capabilities, enabling it to maintain accurate and up-to-date behavioral models that reflect the evolving nature of containerized applications.